EZcap: A Novel Wearable for Real-Time Automated Seizure Detection From EEG Signals

Abstract

Epileptic seizures present a serious danger to the lives of their victims, rendering them unconscious, lacking control, and may even result in death only a few seconds after onset. This gives rise to a crucial need for an effective seizure detection method that is fast, accurate, and has the potential for mass market adoption. Kriging methods have a good reputation for high accuracy in spatial prediction, hence, their extensive use in geostatistics. This paper demonstrates the successful application of Kriging methods for an effective seizure detection device in an edge computing environment by modeling the brain as a spatial panorama. We hereby propose a novel wearable for real-time automated seizure detection from EEG signals using three different types of Kriging, namely, Simple Kriging, Ordinary Kriging and Universal Kriging. After multiple experiments with electroencephalogram (EEG) signals obtained from seizure patients as well as those from their healthy counterparts, the results reveal that the three Kriging methods performed very well in accuracy, sensitivity and latency of detection. It was found however, that Simple Kriging outperforms the other Kriging methods with a mean seizure detection latency of 0.81 sec, a perfect specificity, an accuracy of 97.50% and a sensitivity of 94.74%. The results in this paper compare well with other seizure detection models in the literature but their excellent seizure detection latency surpasses the performance of most existing works in seizure detection.

Epileptic Seizure Detection Using Brain-Rhythmic Recurrence Biomarkers and ONASNet-Based Transfer Learning

“…Manually visual inspection of epileptiform abnormalities in EEG signals is time-consuming and may miss meaningful features behind the observed time series. Developing automatic EEG monitoring systems for epileptic patients is significant for accurate diagnosis of epilepsy, early warning of seizures, and targeted assistance in clinical treatments [4]- [6], such as deploying invasive wearable devices [7].…”

Minimizing artifact-induced false-alarms for seizure detection in wearable EEG devices with gradient-boosted tree classifiers

“…Sayeed et al 11 and Olokodana et al 12 present representative examples of seizure detection frameworks based on statistical feature extraction. However, both papers rely on an intracranial setup that requires invasive brain surgery and is therefore not applicable to wearable surface EEG systems.…”

A Low Complexity Estimation Method of Entropy for Real-Time Seizure Detection

“…Therefore, it is necessary to develop a seizure detection system to increase the efficiency and accuracy of diagnosis. Early treatment to reduce the mortality rate caused by sudden unexpected death in epilepsy (SUDEP) [3].…”